How AI assistants find travel availability
Whether a room, a seat or a table is free, and at what price, changes by the minute and depends on the dates, the number of people and the options chosen. An assistant cannot know it in advance: it has to ask someone who does.
Three places an assistant can look
What it learned in training holds no availability, and its prices are out of date. A web search finds pages written for people: many show availability only after dates are entered, and search results are snapshots. A direct call — to an MCP server or an API run by the provider — sends a structured question (place or property, arrival and departure dates, guests, rooms) and gets a structured answer back.
What a good answer contains
The total price for the whole stay or trip, with taxes and fees, not only a nightly or base rate; the currency; what is included; the cancellation terms; and when the answer was produced, because prices and availability expire. An assistant that has these can compare like with like.
Where the booking happens
With most travel services that assistants can call, the assistant searches and compares, and the booking and payment are completed on the provider's own site, where the traveller sees the final terms before paying.
For travel providers
Publish the availability search as an MCP server or a documented API, say so in an llms.txt at the site's root, and list the server in the MCP Registry so that catalogues — this directory among them — can find it.
Travel services and servers · Maps and transport · Browse the directory →
More: What is an MCP server? · What is llms.txt? · How AI assistants find and book travel · How listings are verified · MCP transports: stdio, Streamable HTTP and SSE · Remote or local: which kind of MCP server to use · How to add an MCP server to an AI assistant · What is the official MCP Registry? · What goes in an llms.txt file · OpenAPI for AI agents · Before you connect an MCP server: a checklist